Presentation on theme: "UK hours measures from different sources – a comparison Mari Kangasniemi."— Presentation transcript:
UK hours measures from different sources – a comparison Mari Kangasniemi
Sources of hours of work in the UK Two main sources: Labour Force Survey and New Earnings Survey LFS: a (quarterly) survey of households living in private addresses, available from 1984 NES: 1% sample of employees in employment, based on employers payroll records, available from 1975
How hours measures differ LFS total hours measure includes unpaid overtime (but our comparisons of LFS measures with and without it show this is not of large magnitude) and is based on employees own account NES excludes those who were affected by absence Timing of the reference week: LFS conducted around the year, NES in April NES has a larger sample size of individuals (LFS rotating panel) Because NES only includes employees, we cannot make comparisons for self employed
Some comparisons Correction for weeks worked: we estimated the ratio of actual hours worked and usual hours worked from LFS and used this to correct NES for absences for 1996-2001 Compared trends (rather than levels) in annual hours 1996-2001 (will be updated later for other years) 1996=100 in EUKLEMS industries In some industries NES clearly smoother (e.g. Food, Pulp and Paper) but for some others different trends (e.g. Retail).
Some summary statistics show that in most (52) industries NES average hours have lower standard deviation and smaller max/min ratio For LFS std. dev and ratio max/min is slightly more negatively correlated with sample size than for NES
After some aggregation (down to 40 industries, to attain at least 1000 observations in NES) difference between correlations is reduced slightly. NES hours still have lower standard deviation in 26 industries.
Conclusions For some industries NES clearly has less variation, but this not always the case Should we use trends from NES to fix the LFS hours to reduce variation in hours? What about self-employed?